[HBR] The long, SLOW HBR project

[Walt Hutchens]

> I will redo the [20M] coil in the next couple of days.

Done.  Downward warmup drift of ~3 kcs over about an hour.   (Worse than I
thought it was, but this was a better quality measurement.)

3 kcs is awkwardly high: further work seemed necessary.   I rewound the coil
again, with 8 instead of 7 turns but with the turns spaced about 1/16".
The concept is that (with the right spacing) the length expansion of the
coil (which reduces the inductance) can compensate for the diametric
expansion (which increases inductance).

(In general a coil with a bit of spacing needs one or two more turns to hit
the same inductance as one that is closewound.)

The space wound coil got the warmup drift down to an honest (carefully
measured) 2 kcs or so.   Then I noticed that I had a paper cap bypassing the
cathode of the oscillator tube; replacing it with the largest mica I could
lay hands on lowered drift still farther, to about 1.6 kcs.   You can still
buy leaded NP0 ceramics up to 0.01 mfd with 500 WV ratings and I may get one
of those for this job.

(Any reactive part in an oscillator circuit through which RF flows will
affect the frequency.   Thus such parts should be eliminated when possible
and those that remain must be as temperature stable as possible.)

There, for a day or so, I stalled.  Nothing else in the circuit jumped out
as the cause of drift taking place over most of an hour.

This AM I warmed the set up thoroughly with the 20M coils, then switched to
the cold set of 80M coils with previously measured drift.    The drift
pattern changed substantially:  Almost no upward drift in the first 2-3
minutes so that's in the chassis -- the oscillator tube would be a good bet
as it takes a few minutes for emission to stabilize.   This drift could be
reduced if the tank circuit/tube coupling could be further reduced but since
it happens so quickly, this isn't a priority.

(The 6688 is a sharp cutoff pentode with a transconductance of 16,500.  It's
intended for use in radar IFs so it ought to have low interelectrode
capacitances.   It should be a good LO tube, but I haven't looked at the
spec sheets -- only the summary in the RCA manual.  6.3V/300 mA.  Naturally
there is no 150 mA equivalent ...)

After five minutes or so the warm chassis/cold coil combination drifted
downward at about half the rate found when warming up everything at once.
The chassis buckling effect shouldn't be much of this since the chassis was
already warm and I made the coil switch quickly.   However the 5651
regulator tube has a temperature coefficient:  Rising tube temperatures
decrease its output voltage by perhaps 0.1 volt over the range of temps to
be expected during warmup.   I turned the hair dryer on this tube for a few
seconds (then turned it off) and sure enough, the oscillator frequency was a
few hundred cps lower, then gradually went back up over the next few
minutes. 

It's mounted where I had a good space: Near the 117N7 rectifier/2nd audio
tube.   That's obviously not ideal and in a full rebuild, I'd put it in a
cool spot.   It should not be too near the oscillator, however, since RF
fields also affect it.

The other way to go is to run the oscillator with the plate voltage roughly
three times the screen voltage so as to reduce the sensitivity to voltage
variations and -- I hope -- eliminate the need for regulating the voltages.
I will probably try that shortly as it is not hard to do and would not only
eliminate the issue coming from the regulator temperature coefficient but
also deal with the problem the regulator was supposed to solve, namely, the
variation of frequency with AGC action.

That ~0.1 volt changes can affect the LO frequency is not a good sign: It
indicates that the oscillator doesn't have a bunch of gain to spare.   And
indeed when I dropped the oscillator plate (tube screen) voltage by half as
an experiment, the LO stopped at the low end of the tuning range on 20M.

The rest of the drift is in the coil assembly.   I could add some more space
to the windings -- that is, I could wind that coil for a fifth time with
greater turn spacing.   Hey, what's another two hours and 3" of PVC pipe?

Use of one or more temperature compensating caps is a possible final step,
when drift is at a practical minimum.

Perhaps I'm over-obsessing but these questions are interesting.  Anyway I'm
going to try to get a bit better stability before making the 10M coils.

Walt 
KJ4KV